The introduction of a centralized system for recording cattle movements in the UK has provided a framework for network-based models for disease spread. However, there are many types of non-reportable contacts between farms which may play a role in disease spread. The lack of real pathogen data with which to test network models makes it difficult to assess whether reported data adequately captures the risk-potential network between farms and improves the accuracy of disease forecasts. A novel multi-disciplinary approach is described whereby network-based models, built upon reported cattle movements and non-reportable local contacts between study farms, are parameterized using field data on bovine Staphylococcus aureus strains. Reported cattle movements were found to play a role in strain spread between farms, but other contacts via farm visitors were also correlated with strain distribution, suggesting that parameterizing contact networks using cattle-tracing data alone may not adequately capture the disease dynamics.